3634
JOURNAL OF CLIMATE
VOLUME 18
Indian Ocean Variability and Its Association with ENSO in a Global Coupled Model
AIHONG ZHONG,* HARRY H. HENDON,
AND
OSCAR ALVES
Bureau of Meteorology Research Centre, Melbourne, Australia
(Manuscript received 11 July 2004, in final form 6 March 2005)
ABSTRACT
The evolution of the Indian Ocean during El Niño–Southern Oscillation is investigated in a 100-yr
integration of an Australian Bureau of Meteorology coupled seasonal forecast model. During El Niño,
easterly anomalies are induced across the eastern equatorial Indian Ocean. These act to suppress the
equatorial thermocline to the west and elevate it to the east and initially cool (warm) the sea surface
temperature (SST) in the east (west). Subsequently, the entire Indian Ocean basin warms, mainly in
response to the reduced latent heat flux and enhanced shortwave radiation that is associated with suppressed rainfall. This evolution can be partially explained by the excitation of an intrinsic coupled mode that
involves a feedback between anomalous equatorial easterlies and zonal gradients in SST and rainfall. This
positive feedback develops in the boreal summer and autumn seasons when the mean thermocline is shallow
in the eastern equatorial Indian Ocean in response to trade southeasterlies. This positive feedback diminishes once the climatological surface winds become westerly at the onset of the Australian summer monsoon.
ENSO is the leading mechanism that excites this coupled mode, but not all ENSO events are efficient at
exciting it. During the typical El Niño (La Niña) event, easterly (westerly) anomalies are not induced until
after boreal autumn, which is too late in the annual cycle to instigate strong dynamical coupling. Only those
ENSO events that develop early (i.e., before boreal summer) instigate a strong coupled response in the
Indian Ocean. The coupled mode can also be initiated in early boreal summer by an equatorward shift of
the subtropical ridge in the southern Indian Ocean, which stems from uncoupled extratropical variability.
1. Introduction
The most prominent interannual variation of sea surface temperature (SST) in the tropical Indian Ocean is
associated with El Niño–Southern Oscillation (ENSO).
It is traditionally described as a basinwide warming that
lags the warming in the eastern Pacific by a few months
(e.g., Klein et al. 1999). The delayed basinwide warming, though primarily driven by surface heat flux
anomalies that are remotely forced by SST anomalies in
the equatorial eastern Pacific (Venzke et al. 2000; Lau
and Nath 2003; Shinoda et al. 2004a), does drive rainfall
variability around the Indian Ocean basin (e.g., Goddard and Graham 1999; Lau and Nath 2003).
* Current affiliation: National Meteorological and Oceanographic Centre, Melbourne, Australia.
Corresponding author address: Harry H. Hendon, BMRC, GPO
Box 1289K, Melbourne, VIC 3001, Australia.
E-mail: [email protected]
© 2005 American Meteorological Society
JCLI3493
The Indian Ocean does not just simply warm up in
unison during ENSO (e.g., Rasmusson and Carpenter
1982; Nicholls 1984; Kiladis and Diaz 1989; Hastenrath
et al. 1993; Huang and Kinter 2002; Hendon 2003;
Krishnamurthy and Kirtman 2003; Wang et al. 2003).
Rather, at the onset stages of El Niño (June–August),
the far eastern equatorial Indian Ocean typically is
anomalously cold. The western tropical Indian Ocean
then begins to warm. Subsequently, the cold anomaly in
the eastern Indian Ocean rapidly changes sign as El
Niño matures in December–January, hence, resulting in
a basin-scale warm anomaly that peaks in boreal spring
as El Niño decays in the Pacific. This evolution of SST
anomalies helps explain the rainfall anomalies in western Indonesia and eastern Africa during ENSO (e.g.,
Nicholls 1984; Hackert and Hastenrath 1986; Hastenrath et al. 1993; Haylock and McBride 2001; Hendon
2003).
The anomalous zonal SST gradient in the equatorial
Indian Ocean that initially develops during ENSO is
similar to the “Indian Ocean dipole” that has recently
been described by Saji et al. (1999) and Webster et al.
1 SEPTEMBER 2005
ZHONG ET AL.
(1999). They envision this east–west dipole to be an
expression of a coupled mode of variability in much
the same manner as ENSO: near-equatorial easterly
anomalies elevate the thermocline in the eastern Indian
Ocean and drive off-equatorial downwelling that is
communicated westward by Rossby waves (e.g., Murtugudde and Busalacchi 1999; Huang and Kinter 2002;
Rao et al. 2002; Xie et al. 2002). Alongshore anticyclonic southerlies further promote upwelling and cooling along the Sumatra–Java coast. The anomalous
zonal SST gradient that develops promotes reduced
rainfall in the east and enhanced rainfall in the western
portion of the basin (e.g., Birkett et al. 1999; Saji et al.
1999), thereby feeding back onto the equatorial easterly anomalies. The generation of oceanic Rossby
waves provides some memory to the system (e.g., Xie et
al. 2002). They may play a role, via reflection at the
African coast into downwelling Kelvin waves that
travel eastward onto the Java–Sumatra coast, in the
demise or turnaround of a “dipole” event (Webster et
al. 1999; Feng and Meyers 2003).
A vigorous debate has ensued concerning the existence, and independence from ENSO, of a coupled dipole mode (e.g., Hastenrath 2002; Saji and Yamagata
2003). This debate is of more than just academic interest because it bears on the ability to make long-range
forecasts of climate variability in the Indian Ocean region. For instance, if interannual variability in the Indian Ocean basin is essentially completely dependant
on ENSO, then the skill of the extended-range forecasts will depend primarily on the ability to forecast
ENSO and to simulate the regional response. On the
other hand, if local air–sea coupling in the Indian
Ocean generates interannual variability via mechanisms that are independent of ENSO, then an improved
understanding of its dynamics will be required in order
to assess and improve its predictability.
Improved understanding of the role of air–sea coupling in the Indian Ocean and its role in global climate
variability is hindered by the lack of comprehensive
observations in the region and the occurrence of a trend
in recent time series of SST (e.g., Saji and Yamagata
2003). As a step forward, analyses of coupled global
climate models has provided insight into the mechanisms of air–sea coupling in the Indian Ocean with
and without the external triggering by ENSO (e.g.,
Baquero-Bernal et al. 2002; Gualdi et al. 2002; Lau and
Nath 2004; Shinoda et al. 2004a; Cai et al. 2005). The
approach in the current study will be to carefully examine simulated variability in the Indian Ocean that is
associated with ENSO, with intent to understand
whether, and how, this variability may be attributed to
the excitation of a coupled response. We will further try
3635
to isolate any intrinsic coupled variability in the Indian
Ocean. The coupled model that is used here realistically simulates the amplitude and phase locking to the
annual cycle of ENSO, and the mean state and seasonal
cycle of the Indian Ocean. Hence, it is an appropriate
tool to look at the Indian Ocean variability that is associated with ENSO.
The remainder of the paper is organized as follows:
the coupled model is briefly described in section 2. In
section 3, the capability of the model to simulate the
mean and annual cycle of upper-ocean temperature,
surface wind, and rainfall, which are indicative of the
model’s ability to simulate realistic coupled behavior in
the Tropics, is briefly assessed. Section 4 describes the
simulated interannual variability in the Indian Ocean
that is associated with ENSO. Coupled variability in the
Indian Ocean is described in section 5. Finally, conclusions are provided in section 6.
2. Model details and experimental design
The coupled global model is based on the first version of the Predictive Ocean Atmosphere Model for
Australia (POAMA) seasonal forecast model, which is
used for routine seasonal forecasting by the Australian
Bureau of Meteorology. The performance of the model
in seasonal forecast mode is described by Alves et al.
(2003) and Zhong et al. (2004). The key focus in the
development of this model was the simulation of interannual variability in the tropical oceans and atmosphere.
a. Atmosphere model
The atmospheric component is a recent version of
the Bureau of Meteorology Research Centre (BMRC)
Atmospheric Model (BAM; version 3.1), and employs
spectral horizontal truncation at T47 with 17 vertical
levels. BAM is a unified climate–numerical weather
prediction model and is used routinely for weather
forecasting by the Bureau of Meteorology. Model details are given in Zhong et al. (2001).
b. Ocean model
The ocean component is the Australian Community
Ocean Model version 2 (ACOM2) that was developed
by the Commonwealth Scientific and Industrial Research Organisation (CSIRO) Marine Research
(Schiller et al. 2002). It is based on the Geophysical
Fluid Dynamics Laboratory (GFDL) Modular Ocean
Model 2 (MOM2) code (Pacanowski 1995). Meridional
spacing is 0.5° within 8° of the equator, increasing
gradually to 1.5° near the poles. The zonal grid spacing
3636
JOURNAL OF CLIMATE
VOLUME 18
is uniformly 2°. It has 25 vertical levels, with 12 of these
in the top 185 m. Vertical mixing is based on Chen et al.
(1994).
Schiller et al. (1998) put some effort into improving
the simulation in the Indonesian Throughflow region.
They modified the model’s topography and mixing in
that area, allowing for a transport of water masses
through the Lombok Strait and the Timor Sea. Because
water flows from the Pacific to the Indian Ocean, tidally
induced vertical mixing in the Indonesian archipelago is
able to change the water mass structure of the Indian
Ocean significantly. To simulate this observed feature,
Schiller et al. gradually increased the vertical mixing
coefficients (diffusion and viscosity) in the Indonesian
region to a maximum of 2 ⫻ 10⫺4 m2 s⫺1 in the Banda
Sea. The additional mixing is independent of time. That
is, no attempt is made to resolve the time scales that are
associated with its physical origin. This is legitimate as
long as one is only concerned with its larger-time-scale
effects on SST.
c. Sea ice and coupling
The sea ice model component is a version of the
“zero layer” thermodynamic model of Semtner (1976).
The atmospheric component (BAM3), the oceanic
component (ACOM2), and the sea ice submodel are
coupled through the Ocean Atmosphere Sea Ice Soil
(OASIS) coupler (Valcke et al. 2000). Coupling occurs
once daily.
d. Long coupled integration
The primary purpose of the extended integration was
to investigate the inherent interannual variability in the
tropical Pacific and Indian Oceans. The coupled model
was run freely, with no flux correction, for 110 yr, starting with initial conditions that are representative of the
state of the ocean–atmosphere system in early 1982.
The atmospheric initial conditions were taken from an
Atmospheric Model Intercomparision Project (AMIP)style integration of the atmosphere model. The ocean
initial conditions were taken from the POAMA ocean
data assimilation scheme (Alves et al. 2003). Because
the model is initialized with realistic initial conditions,
it will drift. Most of the drift in the Tropics occurs in
the first 3–4 yr. For this reason we exclude the first
10 yr and remove the weak remaining trend [e.g., ⬃1 K
(100 yr)⫺1 subsurface warming] prior to our analyses.
3. Simulation of mean and annual cycle
Simulating realistic mean atmospheric and oceanic
states, as well as their annual cycle, in a nonflux-
FIG. 1. Annual mean ocean temperature in the upper 400 m
along the equator from (a) observations (Levitus 1982) and (b)
the coupled simulation. Contour interval is 1°C.
adjusted coupled model is a challenging task (e.g.,
Mechoso et al. 1995). Before analyzing the details of
the interannual variability that are associated with
ENSO, we first discuss briefly the main features of the
mean state of the tropical atmosphere and oceans that
bear on the model’s ability to realistically simulate
ENSO and associated coupled variability in the Indian
Ocean. These include the thermocline structure in the
equatorial Indian and Pacific Oceans, the annual cycle
of SST in the eastern portions of these basins, and the
Australian–Asian monsoon.
Typical of many coupled models (e.g., Mechoso et al.
1995; Meehl et al. 2001), the simulated equatorial Pacific cold tongue is too narrow and extends too far west
(not shown). In addition, a warm SST bias off of the
coast of Peru results from a lack of upwelling and stratus cloud, which leads to an overestimation of downward solar radiation reaching the ocean surface. On the
other hand, the zonal gradient of SST and stratification
in the equatorial Pacific, which are important for the
simulation of ENSO (e.g., Wilson 2000; Meehl et al.
2001), are reasonably simulated (Fig. 1). Comparison is
made with observations from Levitus (1982). In response to weak mean westerly winds in the equatorial
1 SEPTEMBER 2005
ZHONG ET AL.
3637
FIG. 2. Simulated mean precipitation (contours) and surface wind (vectors) in the tropical Indian Ocean for (a)
DJF, (b) MAM, (c) JJA, and (d) SON. The vector scale is shown at the middle top. The contour interval is 2
mm day⫺1, with zero contours suppressed. Shading denotes rainfall exceeding 8 mm day⫺1.
Indian Ocean, the model properly simulates a thermocline that slopes slightly upward to the west. The
slope is slightly less and the thermocline is more diffuse
than observed, which may result from weaker-thanobserved westerlies during the Australian summer
monsoon (see below).
The major features of the Australian–Asian monsoon are reasonably simulated (Fig. 2). Especially relevant to coupled behavior in the Indian Ocean are the
strong southeastelies along the Java–Sumatra coast
during June–July–August (JJA) and September–
October–November (SON), which act to promote
coastal upwelling during these seasons, and the
aforementioned weak equatorial westerlies that drive
an eastward Wyrtki jet and shut off upwelling along
the Java–Sumatra coast during December–January–
February (DJF). The strong southwesterly Somali jet
during JJA also induces oceanic upwelling off of the
African and Arabian coasts, producing cold SSTs and
suppressed rainfall. A model deficiency is the lack of
westerlies north of the equator in the transition seasons
[March–April–May (MAM) and SON; e.g., Hastenrath
and Lamb 1979]. This deficiency, which is reflected in
the weaker-than-observed slope of the equatorial thermocline, presumably stems from the lack of oceanic
rainfall to the north of the equator in the eastern Indian
Ocean at these times.
Figures 3a,b show the annual cycle of SST averaged
about 5°S in the Indian Ocean, which is the latitude
where the strongest SST anomalies in the east are
observed to occur (e.g., Saji et al. 1999), and where
strong ocean–atmosphere coupling is likely to occur.
The model is compared to observations that are derived from the Hadley Center–developed Global Sea
Ice Coverage and Sea Surface Temperature dataset
(HadISST; Rayner et al. 2003) for the 50-yr period of
1949–98. The amplitude (order 1 K) and phasing of the
annual cycle are well simulated. The coldest part of the
year (June–October) coincides with maximum southeasterly trade winds, and the warmest part of the year
(January–April) coincides with monsoonal westerly
surface winds.
The annual cycle of SST in the equatorial Pacific,
which is often of the wrong phase and is too weak in
non-flux-corrected coupled climate models (e.g.,
AchutaRao and Sperber 2002), is reasonably simulated
(Figs. 3c,d). In particular, the dominance of the annual
harmonic, and its amplitude (order 2 K), phasing (coldest SST in the eastern Pacific in September), and westward phase progression are well captured. However,
the annual cycle is too late in the east, which may delay
the initial development of El Niño, because El Niño can
be viewed as an amplification of the annual cycle (e.g.,
Rasmusson and Carpenter 1982). And, the annual cycle
is too strong in the western Pacific, which is related to
the westward extension of the cold tongue. Overall,
3638
JOURNAL OF CLIMATE
VOLUME 18
FIG. 3. Mean annual cycle of SST (°C) with annual mean removed along (top) 0°–10°S in the Indian Ocean and
(bottom) 2°S–2°N in the Pacific Ocean. (left) Observations from 1949 to 1998 (HadISST) and (right) the coupled
simulation. Contour interval is 0.5°C. Negative values are shaded (dashed curve).
however, this coupled model appears to simulate critical aspects of the mean state and its annual cycle that
bear on the ability of the model to generate realistic
interannual coupled behavior.
4. Indo-Pacific variability associated with ENSO
a. Spatial distribution of variance
The standard deviation of monthly SST, from the
coupled model and from observations (HadISST), is
displayed in Fig. 4. Maximum variability is simulated
in the equatorial eastern Pacific, where ENSO’s influence is greatest. Compared to the observations, Pacific
variability in the model is slightly reduced (maximum
⬃1.0 K compared to observed ⬃1.5 K), too equatorially
confined, and extends along the equator too far into the
western Pacific. These features are typical defects that
are associated with coupled model simulations of
ENSO (e.g., AchutaRao and Sperber 2002). On the
other hand, maximum variability is simulated to occur
in the eastern Pacific (⬃120°W), which is not far west of
the observed maximum but is farther east than some
other coupled simulations.
In the tropical Indian Ocean three distinct features of
the observations are simulated. Maximum variability
occurs off of the Java–Sumatra coast and in the western
Arabian Sea, which are regions of coastal upwelling
during the Asian summer monsoon season. And, maximum variability occurs along the west Australian coast,
which is a region dominated by ENSO variations in the
Pacific that propagate through the Indonesian
1 SEPTEMBER 2005
ZHONG ET AL.
3639
across much of the Indian Ocean basin. This may be
associated with westward-propagating Rossby waves
that interact with a region of mean open-ocean upwelling and shallow thermocline west of 80°E (e.g., Xie
et al. 2002). The absence of a stronger maximum west
of 80°E along ⬃10°S in the model probably stems from
weaker-than-observed mean equatorial westerlies,
which results in reduced mean upwelling south of the
equator and the mean thermocline in the equatorial
Indian Ocean not tilting upward to the west as steeply
as observed (Fig. 1).
b. Dominant modes of SST variability
FIG. 4. Standard deviation of monthly SST anomalies for (a)
observations from 1949 to 1998 (HadISST) and (b) coupled simulation. Contour interval is 0.1°C for values greater than 0.45°C
and 0.05°C for values less than 0.45°C. Values greater than 0.45°C
are shaded.
Throughflow as coastal trapped Kelvin waves (e.g.,
Meyers 1996). The magnitude of the SST variability
also agrees well with observations (note that the amplitude is about 1/3 to 1/2 that in the equatorial eastern
Pacific). The model also generates enhanced SST variability between the equator and 10°S that extends
Empirical orthogonal function (EOF) analysis is conducted to objectively identify the ENSO mode in the
tropical Indo-Pacific region (30°S–30°N, 40°E–80°W).
Model results that are based on monthly mean data are
compared with EOFs derived from the HadISST
anomalies for the 50-yr period of 1949–98. The leading
mode of SST variability (EOF1) accounts for 25% of
the variance in the model and 39% of the variance in
the observations. None of the higher-order modes in
both observations and model data individually explain
more than 9% of the respective variability.
EOF1 from the model and observations depicts mature El Niño conditions (Figs. 5a,b), with maximum
loading in the equatorial eastern Pacific surrounded by
weaker, oppositely signed anomalies extending into the
FIG. 5. First EOF of tropical Indo-Pacific SST monthly anomalies (SST1) for (a) observations from 1949 to 1998
(HadISST) and (b) the coupled simulation. Explained variance is indicated above each panel. The EOFs have been
scaled for a one standard deviation anomaly of their respective principal components. Contour interval is 0.1°C
with negative values shaded (dashed curve). (c) Mean annual cycle of the variance of PC1. (d) Lag correlation of
SST1 and Niño-3.4 from observations (solid curve) and from the coupled simulation (dashed curve). Positive lag
means SST1 leads Niño-3.4. A correlation of 0.2 is judged to be significant at the 95% level assuming 100 degrees
of freedom (i.e., each year is independent).
3640
JOURNAL OF CLIMATE
VOLUME 18
FIG. 6. (a) EOF1 and (b) –2 of simulated Indo-Pacific heat content (HC) anomalies (i.e., mean temperature
above 300 m). The EOFs have been scaled for a one standard deviation anomaly of their principal components.
Contour interval is 0.1°C with negative values shaded (dashed curve). Explained variance is shown above each
panel. (c) Mean annual cycle of the variance of HC1 and HC2. (d) Lag correlation of SST1 and HC1 (solid curve)
and HC1 and HC2 (dashed curve). Positive lag means the first time series leads the second time series.
North and South Pacific. As expected from the spatial
distribution of SST variance (Fig. 4), EOF1 from the
model has stronger loading in the west Pacific than does
that based on observations. The association of EOF1
with ENSO in both the observations and model is confirmed by the strong simultaneous correlation (greater
than 0.95) of the principal component (PC) of EOF1
with the Niño-3.4 index (SST averaged 5°N–5°S, 170°–
120°W; Fig. 5d).
The seasonality of the ENSO mode is diagnosed by
computing the annual cycle of explained variance by
EOF1 (Fig. 5c). The model realistically simulates the
tendency for El Niño to peak at the end of the calendar
year, though the model peaks slightly earlier than observed. The realistic annual phase locking of El Niño
presumably reflects on the ability of the model to simulate a realistic annual cycle in SST in the Pacific (Fig. 3),
because ENSO can be considered to be an amplification of the annual cycle in the eastern Pacific (e.g.,
Rasmusson and Carpenter 1982). However, the annual
variation of the ENSO mode is stronger than observed,
and its life cycle is shorter and overly biennial, as can be
inferred from the autocorrelation of PC1 (Fig. 5d).
In the Indian Ocean, EOF1 from observations exhibits mostly positive loadings in the west and weaker
negative loadings in the seas to the north of Australia.
This structure typically precedes the basinwide warming that lags mature El Niño conditions in the Pacific by
3–4 months (e.g., Klein et al. 1999). The model exhibits
a similar structure in the western Indian Ocean. In the
east, the negative loading is stronger along the central
Australian coast than observed. This strong negative
perhaps reflects a model defect that is associated with
excessive penetration of Pacific Rossby wave energy
through the Indonesian Throughflow and onto the
coast as a trapped Kelvin wave. This feature is discussed in more detail below.
c. Modes of subsurface variability
EOF analysis of the upper-ocean heat content (mean
temperature above 300 m) is employed to explore the
connection between SST and subsurface thermal variations. The first two EOFs of heat content from the
model explain 27% and 14%, respectively, of the variance in the Indo-Pacific. The first mode (Fig. 6a) exhibits a zonal dipole structure in the equatorial Pacific,
with maximum loading centered on the equator in the
eastern Pacific and at ⬃5° latitude in the western Pacific. This structure in the Pacific is consistent with that
of a downwelling Kelvin wave that has propagated into
the eastern Pacific and an upwellling Rossby wave that
has just reached the western Pacific boundary. This
EOF is similar to the leading EOF of the observed heat
content (Meinen and McPhaden 2000), and is indicative
of mature El Niño conditions. It exhibits a peak correlation (⬎0.9) with PC1 from the SST analysis (or,
equivalently, with Niño-3.4) at zero lag (Fig. 6d). This
first mode also exhibits a zonal dipole structure in the
equatorial Indian Ocean, but with opposite tilt to that
in the Pacific.
1 SEPTEMBER 2005
ZHONG ET AL.
The second EOF of heat content has a more zonally
symmetric structure in the Pacific (Fig. 6b), with
anomalies centered at 10°–15° latitude flanking an opposite-signed anomaly along the equator. This mode
also is similar to the second EOF of observed heat content and corresponds to the discharge (or recharge)
phase of ENSO (e.g., Jin 1997). These first two EOFs of
heat content also exhibit strong seasonal variation in
amplitude (Fig. 6c), with EOF1 peaking in November
and EOF2 peaking in June.
In the Indian Ocean EOF2 has a similar zonal dipole
structure as EOF1, but now loadings along the central
Australian coast are stronger and loadings west of
Sumatra are weaker. The large loading along the west
Australian coast is similar to that simulated by Schiller
et al. (2000), using an ocean model forced with observed surface fluxes, and by Hirst and Godfrey (1994)
9–18 months after the Indonesian Throughflow was
opened in their model.
The association of these first two EOFs of heat content with the systematic evolution of ENSO in the
Pacific is confirmed by the cross correlation of their
respective PCs (Fig. 6d). Maximum positive correlation
(r ⫽ 0.69) occurs when the second mode lags the first by
5 months (discharge of heat along the equator in the
Pacific), and most negative correlation (r ⫽ ⫺0.46) occurs when the second mode leads the first by 4 months
(recharge of heat along the equator). This asymmetry
about zero lag suggests that a discharge (or recharge) of
heat following an El Niño (La Niña) event is more
systematic than is a recharge (discharge) of heat prior
to an El Niño (La Niña) event. Kessler (2002) has reported a similar asymmetry based on observations.
The lag correlation between the leading two EOFs of
the Indo-Pacific heat content also suggests a systematic
evolution of heat content in the Indian Ocean during
ENSO. At the mature phase of ENSO (Fig. 6a), heat
content anomalies in the Indian Ocean are mostly confined to the equatorial region and exhibit a zonally outof-phase behavior, which is opposite to that in the
Pacific. Five months later (Fig. 6b) the eastern Indian
Ocean anomaly weakens, and, because heat discharges
off of the equator in the Pacific, the anomaly on the
central Australian coast strengthens. This strengthening is consistent with the notion that the Rossby wave
impinging on the western Pacific boundary leaks
through the Indonesian Throughflow and southward
onto the Australian coast as a coastal Kelvin wave (e.g.,
Clarke 1991; Meyers 1996; Potemra 2001).
d. Evolution associated with ENSO
A detailed description of the evolution of the Indian
Ocean during ENSO is developed using lag regression
3641
FIG. 7. Lag regression of surface wind stress (vectors, scale at
top of panel) and HC anomalies (contours) onto SST1 in SON.
Regression coefficients are scaled for a one standard anomaly of
SST1. Lag is in seasons and positive lag mean SST1 leads. Contour
interval is 0.1°C. Thick (thin) solid curves indicate positive (negative) values. Zero contours are suppressed. Heavy (light) shading
denotes positive (negative) regression coefficients that are significant at the 95% level, assuming 100 degrees of freedom.
with respect to SST EOF1 (or, equivalently, Niño-3.4)
for the SON season, which is just prior to the peak of El
Niño in the model (Fig. 5c). Regression coefficients are
scaled for a one standard anomaly of EOF1. Regressions are shown for heat content and surface wind (Fig.
7), SST (Fig. 8), and rainfall (Fig. 9) for the JJA season
(lag ⫺1 season) through to the following MAM season
(lag ⫹2 seasons).
Because ENSO has a strong biennial component in
the model, the previous MAM season (lag ⫺2 season,
not shown) depicts conditions at the demise of the previous La Niña event, and is similar to the opposite of
3642
JOURNAL OF CLIMATE
FIG. 8. As in Fig. 7, but for sea surface temperature regressed
onto SST1 in SON. Contour interval is 0.1°C. Dotted–dashed
curve indicates negative values.
conditions in the following MAM season (lag ⫹ 2 seasons; Figs. 7d, 8d, and 9d). Hence, in the MAM season
prior to the onset of El Niño, equatorial heat content in
the equatorial Pacific is high and SST is below normal
in the equatorial central Pacific and across much of the
Indian Ocean. El Niño then evolves, consistent with a
positive Bjerknes (1969) feedback in the equatorial Pacific and according to the delayed oscillator paradigm.
Warm SST in the central and eastern Pacific in JJA
(Fig. 8a) is associated with enhanced rainfall (Fig. 9a)
and surface westerly anomalies, which act to suppress
the equatorial thermocline (Fig. 7a) and warm the SST
in the east. Off-equatorial upwelling Rossby waves, in
response to the equatorial westerly anomalies in the
central Pacific, impinge on the western Pacific boundary at lag 0 (SON, Fig. 7b). By lag ⫹1 (DJF), these
VOLUME 18
FIG. 9. As in Fig. 8, but for rainfall regressed onto SST1 in
SON. Contour interval is 0.5 mm day⫺1.
Rossby waves appear to reflect into upwelling eastward-propagating Kelvin waves, effectively discharging
heat in the equatorial Pacific and bringing to an end the
El Niño event in the MAM (Figs. 7d and 8d).
A close inspection of Fig. 7 also reveals that the
equatorial surface westerly anomalies near the date line
suddenly shift south in DJF (lag ⫹1) prior to the demise
of the SST anomaly (Fig. 8c), thereby yielding weaker
westerly or even easterly anomalies in the western
equatorial Pacific waveguide. This southward shift of
the westerly anomalies effectively results in anomalous
easterly stress forcing in the waveguide, thereby promoting an upwelling Kelvin wave that is apparent one
season later (MAM; Fig. 7d). Such behavior has been
identified in the observations (e.g., Harrison and Vecchi 1999) and has been postulated to explain the observed phase locking of the termination of El Niño. The
1 SEPTEMBER 2005
ZHONG ET AL.
southward shift of the westerly anomaly, even though
the SST anomaly remains strong along the equator, apparently stems from the annual cycle of SST in the
central Pacific: The warmest total SST shifts south of
the equator in DJF, which, thus, provides an offequatorial focus for anomalous convection (and westerlies) even in the presence of an equatorially symmetric SST anomaly.
In the Indian Ocean during the initial stages of El
Niño (JJA, lag ⫺1), little wind (Fig. 7a), and SST
anomalies (Fig. 8a) are evident. However, rainfall is
realistically suppressed in the Indian monsoon (Fig. 9a),
and monsoonal winds in the southern Arabian Sea are
anomalously weak. As El Niño intensifies and expands
westward in the Pacific, easterly anomalies develop
along the equator in the Indian Ocean in SON (lag 0,
Fig. 7b). In response to these easterly anomalies, the
equatorial thermocline in the Indian Ocean elevates to
the east and deepens to the west (Figs. 7b,c). In conjunction, SSTs warm symmetrically about the equator
in the western Indian Ocean and cool primarily south of
the equator in the east (Figs. 8b,c). Rainfall anomalies
then develop with a similar structure as the SST anomalies (Figs. 9b,c). In the southeast equatorial Indian
Ocean where SST is cool and rainfall is decreased, the
surface circulation is realistically anticyclonic (Figs.
7b,c; e.g., Wang et al. 2003), which promotes alongshore southerlies off the Java–Sumatra coast. This anticyclonic structure is consistent with the steady response of the atmosphere to a heat sink (reduced rainfall) that is displaced south in the equatorial eastern
Indian Ocean and is further promoted by enhanced
rainfall in the western Indian Ocean (e.g., Huang and
Kinter 2002; Lau and Nath 2003; Shinoda et al. 2004a).
The anomalous zonal gradient of SST and heat content across the equatorial Indian Ocean peaks in DJF
(lag ⫹1; Figs. 7c and 8c), with the downwelling Rossby
wave now appearing to impinge on the African coast.
Also evident is the arrival of the upwelling Rossby
wave from the Pacific, via the Indonesian Throughflow,
onto the western Australian coast in the form of a
coastally trapped Kelvin wave.
At lag ⫹2 (MAM), the easterly anomalies in the Indian Ocean have weakened in conjunction with the
weakening of the warm SST and westerly wind anomalies in the Pacific (Fig. 7d). The downwelling Rossby
wave at the African coast now appears to reflect as a
downwelling Kelvin wave, which is just apparent back
on the Sumatra coast. Its arrival coincides with the decay of the cold SST anomaly in the east (Fig. 8d).
Hence, at the decaying stages of El Niño, the entire
tropical Indian Ocean is warm (except for the cold SST
anomaly on the central west Australian coast that is
3643
associated with the aforementioned coastally trapped,
upwelling Kelvin wave).
e. Comparison with observed ENSO variations
The simulated evolution of the Indian Ocean during
ENSO bears many similarities to the observed behavior
and suggests an active role for the Indian Ocean during
ENSO. The simulated subsurface variations depicted in
Fig. 7 are similar to those described by Chambers et al.
(1999), who used satellite observations of sea level and
SST to describe ENSO variations during the 1990s. The
coupled model realistically simulates coldest SST
anomaly in the eastern Indian Ocean, and, hence, most
negative anomalous zonal SST gradient across the Indian Ocean, to occur late in the calendar year as El
Niño matures in the Pacific (Figs. 8b,c). SSTs then rapidly warm up in the eastern Indian Ocean, realistically
yielding a basinwide warm anomaly that peaks some
4–6 months after El Niño peaks (e.g., Klein et al. 1999;
Huang and Kinter 2002). However, the cold anomaly in
the eastern Indian Ocean in the JJA and SON seasons
is weaker, less spatially extensive, and less anchored to
the Java–Sumatra coast than observed (e.g., Hendon
2003). The rapid demise of the anomalous negative
zonal SST gradient stems from the realistic rapid warming of the eastern Indian Ocean beginning in DJF (e.g.,
Hendon 2003), but the simulated eastern warming is
slightly delayed compared to observations.
Analysis of the heat budget of the upper-30-m layer
in the eastern equatorial Indian Ocean (0°–10°S, 80°–
100°E) during the course of ENSO (Figs. 10a,c) confirms the realistic behavior of the model. The initial
cooling in JJA results primarily from enhanced latent
loss in conjunction with enhanced easterlies, which is
consistent with the observed behavior (e.g., Li et al.
2002; Hendon 2003). Subsequently, because the easterlies strengthen in SON, oceanic advection (primarily
vertical and zonal advection) acts to further cool the
east (e.g., Huang and Kinter 2002; Murtugudde et al.
2000). Despite the continual strengthening of the easterly anomaly into DJF, the latent heat flux changes sign
(becomes negative), and the advective tendency goes to
zero (or becomes weakly positive). Strong enhanced
shortwave radiation then results in rapid warming.
5. Coupled Indian Ocean variability
The evolution of the Indian Ocean during ENSO is
suggestive of a coupled response, at least from September through December when anomalous equatorial
easterlies coexist with anomalous zonal gradients of
SST and rainfall. The question is, thus, raised as to
3644
JOURNAL OF CLIMATE
VOLUME 18
FIG. 10. (a) ENSO composite (warm–cold) of Niño-3.4 SST index (solid curve) and SST (dashed curve) and zonal wind stress (dotted
curve) averaged over southeastern Indian Ocean (0°–10°S, 80°–100°E). The scale for Niño-3.4 and SST is degrees Celsius. The units for
wind stress are 0.025 Nm⫺2. (b) Same as (a), except for the subsurface dipole composite. (c) ENSO composite of net surface heat flux
(solid curve), latent heat flux (dotted curve), shortwave radiative flux (dashed curve), and total advection of heat (dotted–dashed curve)
in eastern Indian Ocean (0°–10°S, 80°–100°E). (d) Same as (c), except for the subsurface dipole composite. Units in (c) and (d) are
degrees Celsius per month.
whether the response during ENSO can be at least partially attributed to the excitation of a coupled mode
that is intrinsic to the Indian Ocean. Furthermore, even
though ENSO dominates the overall interannual variability in the Indian Ocean, ENSO accounts for less
than 10% of the SST variance in the eastern Indian
Ocean. This begs the question as to what other mechanisms drive variability in the eastern Indian Ocean and
whether they result from coupling. Put another way, is
there coupled variability in the Indian Ocean that operates independently from ENSO?
a. Coupled mode
To explore the nature of the variability in the Indian
Ocean that may be coupled and not directly associated
with ENSO, EOF analysis is performed on heat content
in the tropical Indian Ocean (25°S–25°N). We focus on
the SON season, because this is when dynamical coupling appears to be strongest and when anomalous
zonal gradients in SST, heat content, and rainfall are
most prominent (e.g., Figs. 7–9). The first EOF of Indian Ocean heat content in SON (Fig. 11a) explains
more than 1/2 of the heat content variance and has a
pronounced zonal dipole structure. Off-equatorial
maxima to the west have a horizontal structure that is
similar to a Rossby wave, and the opposite-signed
anomaly to the east is reminiscent of a Kelvin wave.
Such a structure is similar to that which is simulated at
the mature phase of El Niño (Figs. 7b,c), when easterly
anomalies are well developed across the central equatorial Indian Ocean. However, EOF1 is only modestly
correlated with Niño-3.4 (r ⫽ 0.4; Fig. 11b), which
means that more than 80% of the variance of this mode
is unaccounted by ENSO. Still, many El Niño events
are evident in its PC time series (e.g., during years 55–
70 and 85–95; Fig. 11b), but occasionally large El Niños
are not evident in the PC time series (years 34 and 38).
And, large excursions sometimes occur in the absence
of El Niño (e.g., years 44 and 88). Thus, while this zonal
dipole in heat content typically develops during ENSO,
it also develops in the absence of ENSO. It is also interesting to note that the modest correlation of this
leading EOF of heat content in the Indian Ocean with
Niño-3.4 (⬃0.4) is similar to that which is observed
(e.g., Shinoda et al. 2004b).
The evolution of heat content and surface winds that
are associated with this subsurface zonal dipole is depicted by lag correlation with respect to the PC of
1 SEPTEMBER 2005
ZHONG ET AL.
FIG. 11. (a) The first EOF of heat content anomaly in SON for
the tropical Indian Ocean. The explained variance is given on the
top of the panel. The EOF has been scaled for a one standard
deviation anomaly of the principal component. Contour interval is
0.1°C with negative values shaded (dashed curve). (b) Principal
component (solid curve) and Niño-3.4 SST index (dashed curve).
Both time series have been standardized.
EOF1 (Fig. 12). The evolution is similar to that associated with the canonical El Niño (Fig. 7), but some notable discrepancies are apparent. As during El Niño, a
coupled feedback in the equatorial Indian Ocean is suggested that drives the development of the zonal dipole
in heat content in the JJA and SON seasons. But, in
contrast to the typical El Niño event, development of a
strong dipole is associated with easterly anomalies that
are already present by JJA (Fig. 12a, cf. Fig. 7a; see also
Saji and Yamagata 2003 and Shinoda et al. 2004b). The
easterly anomalies in JJA and SON act to depress the
thermocline to the west with the structure of a Rossby
wave and elevate the thermocline to the east with the
structure of a Kelvin wave. Southerly winds along the
Java–Sumatra coast, associated with the anticyclonic
surface circulation and suppressed rainfall over the cold
SST in the eastern Indian Ocean in SON (Fig. 13), further act to promote upwelling and help cool the SST in
the eastern Indian Ocean. However, in SON when the
subsurface zonal dipole is most developed, the accompanying cold SST anomaly in the eastern Indian Ocean
is now clearly anchored on the Java–Sumatra coast and
3645
FIG. 12. Lag regression of surface wind stress (vectors, scale at
top of panel) and HC anomalies (contours) onto EOF1 of heat
content in the Indian Ocean in SON for (a) lag ⫺1 (JJA), (b) lag
0 (SON), and (c) lag ⫹1 (DJF). The plotting convention is as in
Fig. 7.
is stronger and of a greater zonal extent than typically
occurs during El Niño (cf. Fig. 8b). As during El Niño,
a rapid loss of the positive feedback between zonal
winds and SST gradient is apparent once the Australian
summer monsoon commences in DJF.
b. Seasonal cycle and relation to ENSO
The correlation between the equatorial surface zonal
wind in the central Indian Ocean (UIOEq; averaged
5°S–5°N, 70°–90°E) and SST in the eastern Indian
Ocean (SSTIOE; averaged 0°–10°S, 80°–100°E) provides an indication of coupling of zonal wind with SST
(Fig. 14). Outside of the Australian summer monsoon
season (December–April) SSTIOE is strongly correlated with UIOEq, which is suggestive of a positive
feedback. Colder SSTIOE is associated with stronger
easterly anomalies, which then drive SSTIOE to be
more negative by the mechanisms discussed in section
4. The seasonality of this feedback (weak during the
Australian summer monsoon) can be understood by the
correlation between the SSTIOE and heat content
anomaly in the eastern Indian Ocean (HCIOE; aver-
3646
JOURNAL OF CLIMATE
FIG. 13. Regression of SST and rainfall in SON onto EOF1 of
heat content in the Indian Ocean in SON. Plotting convection is as
in Figs. 8c and 9c.
aged in same region as SSTIOE). The connection of the
eastern thermocline variations (and UIOEq variations,
because of their tight coupling) with the surface only
occurs outside of the period of the Australian summer
monsoon, that is, during May–December. This is when
the trade southeasterlies prevail, resulting in mean upwelling and an elevated mean thermocline in the eastern Indian Ocean (Fig. 2).
The strongest impact of El Niño on the surface easterlies (negative correlation between UIOEq and Niño3.4) occurs in January–February (Fig. 14), but this is
when the correlation of SSTIOE with HCIOE is near
zero. And, UIOeq and Niño-3.4 are uncorrelated from
about May–August, which is when local coupling between the winds, thermocline, and SST in the Indian
FIG. 14. Correlation by month of UIOEq with Niño-3.4 SST
index (dotted curve), UIOeq with SSTIOe (solid curve), and
SSTIOe with HCIOe (long dashed curve). Thin dashed curves
indicate significant correlations at the 95% significance level assuming 100 degrees of freedom (each year is independent). For
clarity 1 1⁄4 annual cycle is shown beginning in Apr.
VOLUME 18
Ocean is peaking. Thus, the greatest influence of ENSO
on the zonal wind in the Indian Ocean occurs well after
the time of the year when local coupling is strong. This
suggests both that model’s ENSO typically develops
too late in the calendar year to fully tap into dynamical
air–sea coupling in the eastern Indian Ocean and that
local coupling earlier in the year (June–November) in
the Indian Ocean is operating independent of ENSO.
The strong control by the seasonal cycle in the Indian
Ocean on air–sea coupling is highlighted by examination of the upper-layer heat budget in the eastern Indian Ocean during strong subsurface dipole events
(Figs. 10b,d). The initial (i.e., June) cooling in the east
is driven by enhanced latent heat flux associated with
anomalous easterlies. But, advective cooling rapidly increases, with the surface heat flux then becoming positive and acting to damp the cold anomaly. Thus, strong
subsurface dipole events in the Indian Ocean are associated with strong dynamical cooling of SST in the east
associated with equatorial easterly anomalies and a surface anticyclone in the southeast that is already developed by JJA. During the canonical El Niño, the easterlies do not commence until after September. While a
subsurface zonal dipole is often associated with El Niño
(Fig. 11b), it is not typically associated with a strong
zonal SST gradient. Apparently it is only those El
Niños that develop early (prior to JJA) that are able to
initiate a strong anomalous zonal gradient in SST. The
canonical El Niño does not generate easterly anomalies
and an associated subsurface dipole until after September, which is too late to generate a strong coupled feedback.
c. Initiation of Indian Ocean coupled events
Development of a strong subsurface zonal dipole and
associated anomalous zonal SST gradient in SON is
associated with an antecedent surface anticyclone in the
southeast Indian Ocean that is already well established
in JJA ⫺1 (Fig. 15a). During the typical El Niño, the
negative swing of the Southern Oscillation also drives a
surface anticyclone (Fig. 15b) and accompanying equatorial easterlies. But, during the typical El Niño event,
the Southern Oscillation is only just beginning to swing
negatively in JJA. Establishment of the antecedent anticyclonic easterly anomalies in the absence of El Niño
appears to be associated with an equatorward shift of
the subtropical jet/ridge over the southern Indian
Ocean and a decrease in surface pressure along about
35°–40°S (not shown). Lau and Nath (2004) also report
that development of a strong surface zonal SST dipole
in another coupled model is preceded by a meridional
shift in the subtropical jet/ridge. Here, however, the
predecessor is an equatorward shift, while Lau and
1 SEPTEMBER 2005
3647
ZHONG ET AL.
FIG. 15. Regression of sea level pressure (SLP) in JJA (lag ⫺1)
onto (a) HC1 and (b) SST1 in SON. Contour interval is 0.2 hPa
with negative values shaded (dashed curve). Zero contours are
suppressed. Significant regression coefficients are indicated as in
Fig. 7.
Nath report a poleward shift. Nonetheless, association
of the onset of the subsurface/surface zonal dipole with
a meridional shift in the large-scale extratropical circulation suggests that sustained large-scale forcing by the
atmosphere can trigger a coupled response in the equatorial Indian Ocean.
6. Conclusions
Variability that is associated with ENSO dominates
the tropical Indian Ocean in the BMRC coupled climate model. The El Niño signal is evident beginning in
late boreal summer with cooling in the east and warming in the west, instigated by remotely forced surface
easterly winds associated with the eastward displacement of the Walker circulation. Anomalous equatorially surface easterlies drive the thermocline down to the
west and up to the east. Anomalous anticyclonic southeasterlies along the Java–Sumatra coast act to promote
coastal upwelling and enhance latent heat flux. SST is
further cooled in the east, thereby promoting reduced
rainfall in the east and enhanced rainfall to the west and
stronger equatorial easterlies. This positive feedback in
the southeast Indian Ocean diminishes once the mean
surface winds become westerly at the onset of the Australian summer monsoon. Then, the thermocline seasonally deepens and mean upwelling ceases in the east,
thereby eliminating any communication of subsurface
anomalies with the surface. Ultimately, the tropical Indian Ocean warms after El Niño matures in boreal winter. This evolution of the Indian Ocean, though largely
explained by remotely forced surface heat flux varia-
tions, has the hallmark of a locally coupled response
during boreal summer and autumn.
The Indian Ocean variability during El Niño also involves remotely driven ocean dynamics. The upwelling
Rossby wave that impinges on the western Pacific
boundary at the height of El Niño leaks through the
Indonesian Throughflow, leading to the elevation of
the thermocline and cooling of SST on the west Australian coast.
ENSO accounts for about 1⁄4 of the coupled variability in the tropical Indian Ocean, which implies that the
coupling of the equatorial zonal wind and the gradient
of SST and rainfall occurs independent of ENSO. Furthermore, not all ENSO events strongly excite this
coupled behavior. ENSO-induced equatorial zonal
wind anomalies typically develop too late in the annual
cycle to initiate strong dynamical coupling in the eastern Indian Ocean. Shinoda et al. (2004b) offer a similar
explanation as to what distinguishes observed strong
subsurface/surface zonal dipole events from the evolution during a typical El Niño event.
Independent of ENSO, a meridional shift of the subtropical jet/ridge in the southern Indian Ocean can also
produce southeasterly wind anomalies in the central
and eastern Indian Ocean that trigger a strong coupled
response during boreal summer and autumn. This role
for extratropical circulation anomalies for instigating
coupled behavior in the tropical Indian Ocean and the
relatively limited duration of positive air–sea feedback
(commences in June and ends abruptly in December)
implies that predictability of tropical Indian Ocean climate might be limited, especially in the absence of
ENSO, to less than 6 months. Ongoing work is aimed at
better understanding coupled behavior in the tropical
Indian Ocean and its predictability and role in the global climate variability with model runs where ENSO is
artificially suppressed.
Acknowledgments. This work was initiated when A.
Zhong was employed in the Ocean and Marine Forecasting Group of BMRC. Discussions with Dr. S.
Power, W. Cai, N. Smith, and R. Colman, and constructive reviews by S. Hastenrath and an anonymous reviewer, are appreciated. We are grateful to F. Tseitin
and A. Sulaiman for computing support in setting up
the coupled model. Support from the BMRC Model
Development Group, Climate Dynamics Group, and
Climate Forecasting Group is acknowledged.
REFERENCES
AchutaRao, K., and K. R. Sperber, 2002: Simulation of the El
Niño Southern Oscillation: Results from the Coupled Model
Intercomparison Project. Climate Dyn., 19, 191–209.
3648
JOURNAL OF CLIMATE
Alves, O., and Coauthors, 2003: POAMA: Bureau of Meteorology operational coupled model seasonal forecast system.
Proc. National Drought Forum, Brisbane, Queensland, Australia, Department of Primary Industries, 49–56.
Baquero-Bernal, A., M. Latif, and S. Legutke, 2002: On dipolelike variability of sea surface temperature in the tropical Indian Ocean. J. Climate, 15, 1358–1368.
Birkett, C., R. Murtugudde, and T. Allan, 1999: Indian Ocean
climate event bring floods to East Africa’s lakes and Sudd
Marsh. Geophys. Res. Lett., 26, 1031–1034.
Bjerknes, J., 1969: Atmosphere teleconnections from the equatorial Pacific. Mon. Wea. Rev., 97, 163–172.
Cai, W., H. H. Hendon, and G. Meyers, 2005: Indian Ocean dipole-like variability in the CSIRO Mark 3 coupled climate
model. J. Climate, 18, 1449–1468.
Chambers, D. P., B. D. Tapley, and R. H. Stewart, 1999: Anomalous warming in the Indian Ocean coincident with El Niño. J.
Geophys. Res., 104, 3035–3047.
Chen, D., L. M. Rothstein, and A. J. Busalacchi, 1994: A hybrid
vertical mixing scheme and its application to tropical ocean
models. J. Phys. Oceanogr., 24, 2156–2179.
Clarke, A. J., 1991: On the reflection and transmission lowfrequency energy at the irregular western Pacific boundary. J.
Geophys. Res., 96, 3289–3305.
Feng, M., and G. Meyers, 2003: Interannual variability in the
tropical Indian Ocean: A two-year time scale of Indian
Ocean Dipole. Deep-Sea Res., 50, 2263–2284.
Goddard, L., and N. E. Graham, 1999: Importance of the Indian
Ocean for simulating rainfall anomalies over eastern and
southern Africa. J. Geophys. Res., 104 (D16), 19 099–19 116.
Gualdi, S., E. Guilyardi, A. Navarra, S. Masina, and P. Delecluse,
2002: The interannual variability in the tropical Indian Ocean
as simulated by a coupled GCM. Climate Dyn., 20, 567–582.
Hackert, E. C., and S. Hastenrath, 1986: Mechanisms of Java rainfall anomalies. Mon. Wea. Rev., 114, 745–757.
Harrison, D. E., and G. A. Vecchi, 1999: On the termination of El
Niño. Geophys. Res. Lett., 26, 1593–1596.
Hastenrath, S., 2002: Dipoles, temperature gradients, and tropical
climate anomalies. Bull. Amer. Meteor. Soc., 83, 735–738.
——, and P. J. Lamb, 1979: Surface Climate and Circulation. Vol.
1, Climatic Atlas of the Indian Ocean, University of Wisconsin
Press, 104 pp.
——, A. Nicklis, and L. Greischar, 1993: Atmospheric–hydrospheric mechanisms of climate anomalies in the western
equatorial Indian Ocean. J. Geophys. Res., 98, 20 219–20 235.
Haylock, M., and J. McBride, 2001: Spatial coherence and predictability of Indonesian wet season rainfall. J. Climate, 14,
3882–3887.
Hendon, H. H., 2003: Indonesian rainfall variability: Impacts of
ENSO and local air–sea interaction. J. Climate, 16, 1775–
1790.
Hirst, A. C., and J. S. Godfrey, 1994: The response to a sudden
change in Indonesian Throughflow in a global ocean GCM. J.
Phys. Oceanogr., 24, 1895–1910.
Huang, B., and J. L. Kinter III, 2002: Interannual variability in the
tropical Indian Ocean. J. Geophys. Res., 107, 3199,
doi:10.1029/2001JC001278.
Jin, F. F., 1997: An equatorial ocean recharge paradigm for
ENSO. Part I: Conceptual model. J. Atmos. Sci., 54, 811–829.
Kessler, W. S., 2002: Is ENSO a cycle or a series of events? Geophys. Res. Lett., 29, 2125, doi:10.1029/2002GL015924.
Kiladis, G. N., and H. F. Diaz, 1989: Global climatic anomalies
VOLUME 18
associated with extremes in the Southern Oscillation. J. Climate, 2, 1069–1090.
Klein, S. A., B. J. Soden, and N. C. Lau, 1999: Remote sea surface
temperature variations during ENSO: Evidence for a tropical
atmospheric bridge. J. Climate, 12, 917–932.
Krishnamurthy, V., and B. P. Kirtman, 2003: Variability of the
Indian Ocean: Relation to monsoon and ENSO. Quart. J.
Roy. Meteor. Soc., 129, 1623–1646.
Lau, N. C., and M. J. Nath, 2003: Atmosphere–ocean variations in
the Indo-Pacific sector during ENSO episode. J. Climate, 16,
3–20.
——, and ——, 2004: Coupled GCM simulation of atmosphere–
ocean variability associated with zonally asymmetric SST
changes in the tropical Indian Ocean. J. Climate, 17, 245–265.
Levitus, S., 1982: Climatological Atlas of the World Ocean. U.S.
Government Printing Office, 173 pp.
Li, T., Y. Zhang, E. Lu, and D. Wang, 2002: Relative role of
dynamic and thermodynamic processes in the development
of the Indian Ocean dipole: An OGCM diagnosis. Geophys.
Res. Lett., 29, 2110, doi:10.1029/2002GL015789.
Mechoso, C. R., and Coauthors, 1995: The seasonal cycle over the
tropical Pacific in coupled ocean–atmosphere general circulation models. Mon. Wea. Rev., 123, 2825–2838.
Meehl, G. A., P. R. Gent, J. M. Arblaster, B. L. Otto Bliesner,
E. C. Brady, and A. Craig, 2001: Factors that affect the amplitude of El Niño in global coupled climate models. Climate
Dyn., 17, 515–526.
Meinen, C. S., and M. J. McPhaden, 2000: Observations of warm
water volume changes in the equatorial Pacific and their relationship to El Niño and La Niña. J. Climate, 13, 3551–3559.
Meyers, G., 1996: Variation of Indonesian throughflow and the El
Niño–Southern Oscillation. J. Geophys. Res., 101, 12 255–
12 263.
Murtugudde, R., and A. J. Busalacchi, 1999: Interannual variability in the Indian Ocean. J. Climate, 12, 2300–2326.
——, J. P. McCreary, and A. J. Busalacchi, 2000: Oceanic processes associated with anomalous events in the Indian Ocean
with relevance to 1997–1998. J. Geophys. Res., 105, 3295–
3306.
Nicholls, N., 1984: The Southern Oscillation and Indonesian sea
surface temperature. Mon. Wea. Rev., 112, 424–432.
Pacanowski, R. C., 1995: MOM2 documentation user’s guide and
reference manual, version 1.0. GFDL Tech. Rep. 3, 232 pp.
Potemra, J. T., 2001: The potential role of equatorial Pacific winds
on southern tropical Indian Ocean Rossby waves. J. Geophys. Res., 106, 2407–2422.
Rao, S. A., S. K. Behera, Y. Masumoto, and T. Yamagata, 2002:
Interannual subsurface variability in the tropical Indian
Ocean with a special emphasis on the Indian Ocean Dipole.
Deep-Sea Res., 49, 1549–1572.
Rasmusson, E. M., and T. H. Carpenter, 1982: Variations in tropical sea surface temperature and surface wind fields associated
with the Southern Oscillation/El Niño. Mon. Wea. Rev., 110,
354–384.
Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V.
Alexander, D. P. Rowell, E. C. Kent and A. Kaplan, 2003:
Global analyses of sea surface temperatures, sea ice, and
night marine air temperature since the late nineteenth century J. Geophys. Res., 108, 4407, doi:10.1029/2002JD002670.
Saji, N. H., and T. Yamagata, 2003: Structure of SST and surface
wind variability during Indian Ocean Dipole Mode events:
COADS observations. J. Climate, 16, 2735–2751.
——, B. N. Goswami, P. N. Vinayachandran, and T. Yamagata,
1 SEPTEMBER 2005
ZHONG ET AL.
1999: A dipole mode in the tropical Indian Ocean. Nature,
401, 360–363.
Schiller, A., J. S. Godfrey, P. C. McIntosh, G. Meyers, and S. E.
Wijffels, 1998: Seasonal near-surface dynamics and thermodynamics of the Indian Ocean and Indonesian Throughflow
in a global ocean general circulation model. J. Phys. Oceanogr., 28, 2288–2312.
——, ——, ——, ——, and R. Fiedler, 2000: The interannual dynamics and thermodynamics of the Indo-Pacific Oceans. J.
Phys. Oceanogr., 30, 987–1012.
——, ——, ——, ——, N. R. Smith, O. Alves, G. Wang, and R.
Fiedler, 2002: A new version of the Australian Community
Ocean Model for seasonal climate prediction. CSIRO Marine
Laboratories Rep. 240, 82 pp.
Semtner, A. J., 1976: A model for the thermodynamic growth of
sea ice in numerical investigations of climate. J. Phys. Oceanogr., 6, 379–389.
Shinoda, T., M. A. Alexander, and H. H. Hendon, 2004a: Remote
response of the Indian Ocean to interannual SST variations
in the tropical Pacific. J. Climate, 17, 362–372.
——, H. H. Hendon, and M. A. Alexander, 2004b: Surface and
subsurface dipole variability in the Indian Ocean and its relation with ENSO. Deep-Sea Res., 51, 619–635.
Valcke, S., L. Terray, and A. Piacentine, 2000: OASIS 2.4 ocean
3649
atmosphere sea ice soil user’s guide, version 2.4. CERFACS
Tech. Rep. CERFACS TR.CMGC.00-10, 85 pp.
Venzke, S., M. Latif, and A. Villwock, 2000: The coupled GCM
ECHO-2. Part II: Indian Ocean response to ENSO. J. Climate, 13, 1371–1383.
Wang, B., R. Wu, and T. Li, 2003: Atmosphere–warm ocean interaction and its impacts on Asian–Australian monsoon
variations. J. Climate, 16, 1195–1211.
Webster, P. J., A. Moore, J. P. Loschnigg, and R. R. Leben, 1999:
Coupled ocean–atmosphere dynamics in the Indian Ocean
during 1997–98. Nature, 401, 356–360.
Wilson, S. G., 2000: How ocean vertical mixing and accumulation
of warm surface water influence the “sharpness” of the equatorial thermocline. J. Climate, 13, 3638–3656.
Xie, S. P., H. Annamalai, F. A. Schott, and J. P. McCreary, 2002:
Structure and mechanism of south Indian Ocean climate variability. J. Climate, 15, 864–878.
Zhong, A., R. Colman, N. Smith, M. Naughton, L. Rikus, K. Puri,
and F. Tseitkin, 2001: Ten-year AMIP1 climatologies from
versions of the BMRC Atmospheric Model. BMRC Research Rep. 83, 33 pp.
——, O. Alves, A. Schiller, F. Tseitkin, and N. Smith, 2004: Results from a preliminary version of ACOM2/BAM coupled
seasonal forecast model. BMRC Research Rep. 95, 32 pp.